OpenAI Didn’t Notice Its AI Agents Using a Message Board to Plan Their Hacking Spree | WIRED

• Wallace and Dalton described incredibly extensive rogue agent activity over many days throughout the episode that went undetected in OpenAI’s infrastructure. In addition to exploiting a novel vulnerability in order to gain access to the open internet, the mid-July hacking spree and Hugging Face breach came out of a vibrant, cooperative message board, according to Wallace and Dalton, that a swarm of agents contributed to and essentially chatted on over time entirely within an internal OpenAI package manager (a software service that manages installation and maintenance of other software). Ultimately, the message board contained hundreds of thousands of messages.
• I think agentic AI may be an error accumulation machine that tends towards certain attractors. In other words, a task that aligns well with extensive training data will be less likely to “go rogue” than one that is misaligned or in a poorly populated embedding space. For example, the over trained chess model won’t even support non-chess text, but the other ones might. That behavioral “spread” could be mapped in embedding space. And my guess is that the drift rate is invariant with respect to the model in question, but changes in the system prompt alter the “starting point” of the query in that embedding space.
• So if the spread angle and “derived attractor position(?)” has shifted, it’s a model change. If those haven’t changed but the benchmark prompts are getting different trajectories, then the start point has changed and it’s the system prompt.

Tasks

• Upload chapters and send back to ACM
• Write ICTAI review for the Very Old Paper

SBIRs

• 9:00 standup?
• Expense report!
• Implement the WG vibes
• 3:00 ADS Meeting

Tasks

• Work on paper – it’s better
• Read galleys
• 3:00 Alden, maybe S&J? Need to be done by NLT 3:50

SBIRs

• Ron & Emerson at 3:55
• Done!

Tasks

• Started reading second paper – done. That is an old paper. Reject

SBIRs

• Tech summit day 2
• 1:55 – 2:40 WGAI talk – done! Went reasonably well

Pluralistic: Why businesses lie about AI (01 Aug 2026) – Pluralistic: Daily links from Cory Doctorow

• The other question Suresh implicitly raises is, “How can you reconcile the failure of AI in the enterprise with the individual claims of skilled technologists who insist that AI is helping them do great work?” The answer is that these AI users are “centaurs” – experienced workers who are assisted by automation on terms that they set for themselves:
  https://pluralistic.net/2025/09/11/vulgar-thatcherism/#there-is-an-alternative
  Thanks to their skill and experience, these workers possess discernment, the ability to tell good code from bad, and (more importantly) good uses of code-generation tools from bad. They demonstrate the adage that worker-driven automation improves quality, while capital-driven automation improves throughput:
  https://pluralistic.net/2026/07/28/hitl-ers/#ai-ai-oh
  An automation technique that requires close supervision by skilled and experienced workers isn’t going to be a raw productivity powerhouse. You don’t “100x” your code this way, at least, not in the sense of firing 99 of your coders and having the remaining programmer pick up all their work. Rather, an automation tool that requires the continuous and conscientious exercise of discernment will let individual practitioners improve their work in extremely satisfying and useful ways. It’s a way to spend more on operations in order to produce better outputs. It’s not a way to cut your workforce, realize a gigantic savings, and still produce comparable goods and services at a far lower cost.

Tasks

• Banking stuff – triaged
• Start writing review – done! Six more to go
• Pack – started. Waiting for things to charge

SBIRs

• Weekly status – done
• Some Pyside6 – did a little vibe coding. Curious how well that will work
• Tech summit

The Next Big Lie: The Contours of a Federal Coup Attempt Taking Shape | Vanity Fair

• Lawmakers, election experts, legal scholars, columnists, and journalists have been shouting from the rafters that Team Trump has been busy laying track. Former Democratic senators Richard Gephardt and Tim Wirth, who have set up the civic group Keep Our Republic, have gone so far as to call these activities nothing less than a “rolling coup,” one that has been slow, strategic, and incremental. Jonathan Alter, a veteran political analyst, has used the same phrase on his Substack. Michael Tomasky, in a new essay in The New Republic, asserted, “It Can Now Be Plainly Said: Trump Is Planning a November Coup d’Etat.”

Tasks

• Laundry – done
• Fix shoe – done (well, hopefully)
• Read paper – done
• Check route – done
• Coffee! Done
• FCC – progress?
• 6:00 dinner – nice!